1 files changed, 44 insertions, 0 deletions
diff --git a/findurls.go b/findurls.go
new file mode 100644
index 0000000..6bc95d7
--- /dev/null
+++ b/findurls.go
@@ -0,0 +1,44 @@
+package main
+
+import (
+	"log"
+	"net/url"
+
+	"golang.org/x/net/html"
+	"golang.org/x/net/html/atom"
+)
+
+// findURLs combs htmlDoc for anchor-tag URLs. The refURL parameter is
+// used to resolve each found URL into an absolute URL.
+func findURLs(refURL url.URL, htmlDoc *html.Node) []url.URL {
+	// Used for collecting all the URLs we find.
+	var rawURLs []url.URL
+
+	for node := range htmlDoc.Descendants() {
+		if node.Type == html.ElementNode {
+			switch node.DataAtom {
+			case atom.A:
+				for _, attr := range node.Attr {
+					if attr.Key == "href" {
+						subURL, err := refURL.Parse(attr.Val)
+						if err != nil {
+							log.Printf("bad URL: %s", attr.Val)
+							continue
+						}
+
+						// Filter external
+						// URLs from the final
+						// result.
+						if subURL.Hostname() != refURL.Hostname() {
+							continue
+						}
+
+						rawURLs = append(rawURLs, *subURL)
+					}
+				}
+			}
+		}
+	}
+
+	return rawURLs
+}